![Page 1: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/1.jpg)
Helen Mayberg MDCenter for Advanced Circuit TherapeuticsIcahn School of Medicine at Mount Sinai
New York
October 15, 2019
DBS for Treatment-Resistant Depression: a (5 year) Progress Report
![Page 2: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/2.jpg)
Off-Label Use of Devices: Donated DBS electrodes/pulse generators 1. Medtronic Inc. (Toronto, Emory, MSSM)2. Abbott Labs/St. Jude Medical, Inc (Emory)
Patent: US2005/0033379A1 (Andres Lozano, co-inventor)issued March 2008, Abbott Labs, assignee
Consultant: Abbott Labs
NARSAD Distinguished Investigator Award 2002BBRF Webinar 2014 Today: 5 year update
Disclosures
![Page 3: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/3.jpg)
R Gross P Holtzheimer S Garlow P Riva-Posse A Crowell R Hershenberg S Quinn L Denison
Imaging Electrophysiology
Psychiatry PsychotherapyNeurosurgery Patient Coordination
K Choi J Rajendra A Waters O Smart V Tiruvadi A Veerakumar M Sendi S Alagapan
DBS Biometricsand Mechanisms
Modeling,Behavioral Biometrics
C McIntyre B Howell D Obatusin T Denison S Nemati S Hamati C Inman. M KelleyCase Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatisticsmodeling modeling Engineering
ClinicalImplant
Programming
NIMH 1R01MH102238, 1R01MH106173, BRAIN UH3NS103550 FDA IDE: G060028, G130107 (PI: HM)Clinicaltrials.gov ID#: NCT00367003, NCT01984710
Emory Depression DBS Team
![Page 4: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/4.jpg)
Movement Disorders
Addictive DisordersSeizures, Memory
hippocampusM. bodies
Fornix
Circuit of Papez
Obsessive-CompulsiveDisorders (OCD, TS)
Axiom 2019neuropsychiatric disorders are circuitopathies
PDET
Dystonia(TS)
(Paralysis)
Mood Disorders(MDD, PTSD, anxiety)
Opiatescocainealcoholobesity
AlzheimersEnhancement
TLE
Chronic Pain
![Page 5: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/5.jpg)
Focal Modulation of Disease Circuitsgeneral approach (invasive/non-invasive)
• WHY? (define need)
• WHERE to stimulate (critical node)
• WHAT should happen (target engagement, endpoint)
• WHO to stimulate (patient selection biomarker)
• HOW to stimulate (intermittent, continuous, closed loop)
Goal - Match Target to disease, symptom, patient
- Devise personalize algorithm to optimize response
![Page 6: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/6.jpg)
electrode
IPG: implantablepulse generator
MRI/CT Guided targeting
Stereotaxic Implantation+/- awake, recording, testing
DBS system in situdisease specified locationchronic continuous stim
DBS 101: Basic ProcedureTarget and modulate a neural circuit
Equipment
HD
RS
17 s
core
![Page 7: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/7.jpg)
Treatments are available, but not always effective• 10% become treatment resistant over time• few options if fail ECT
Rationale for Neuromodulation as a Potential Strategy• advances in functional neurosurgery and imaging (essential)• experience in Parkinson’s disease (naïve but a start)
DBS for Depression: Motivation 2001Why?
MoodInterest
ActivitiesWeightSleep
ActivityEnergy
ConcentrationGuilt
Suicide
![Page 8: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/8.jpg)
DBS for TRDWhat are we trying to treat?
nearly immobilized and in a tranceof supreme discomfort…
William Styron. Darkness Visible 1991 (2004)
ActionIntention
Mood
“A gnawing agony; a painful self-loathing that consumes all your energy and attention…”
“Can’t get away from inside yourself…”
What might recovery look like?can move; be without pain?
Return of agency?
![Page 9: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/9.jpg)
Toronto: Pilot Proof of principle
Cg25 Cg25
psychic painneg mood
depression recovery
cc
mid-SCC
genu ac
Target SCC WM 130Hz 90us 4V
Baseline Ham17=27+2
mCC
vCdsncg25 hth
6m ChangeHam17=7.8+3
Cg25
mF10
oF11
MCC
bs
mF10
mechanismCBF PET
Rationale GoalOutcome HAM-17 (Classic Dep Rating Scale)
Method
Hypothesis: blocking BA25 willalso change regions connected to it
Proof-of-Principle Pilot Study: 6 TRD patients6-month open-label DBS, 1st pt 2003, published 2005
Simple Minded Approachunambiguous, go-no-go outcomes
TRD pts >4yr CE, >4 Rx, fail ECT, Ham>20
![Page 10: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/10.jpg)
Phase 2: Extension, Replication, MaintenanceExpansion to other sites 2008-2012
Lozano Biol Psych 2008 Holtzheimer et al. Arch Gen Psych 2012
05
10152025
Base 1 2 3 4 5 6 7 8 9 10 11 12
Tota
l HAM
-D 1
7 Sc
ore
All time points p < 0.001
Months Post-DBS
Resp=60%; Resp=55%
months after implant
Toronto 20 patients: 6 mo initial, 1yr f/u
BL sh 1m 2 3 4 5 6 7 8 9 10 11 12 2y
24
18
12
6
0
Emory n=17 1st pt 2008 Bipolar-2 = MDD
No changein meds for 6 months Last f/u: 12/14 (80%R) T0=2015
3 explanted, 11 new cohort
Resp 41% 36% 90% Rem 18% 6% 58%
1 mo placebo single blind, 18m open6m open label, 6m continuation
avg=42 mo
ITOC
Resp 62.5% 46.2% 75% 64%Rem 18.8% 15.4% 50% 42%
years after implant
Toronto n=20 Long Term f/u: 3-6 yrs
HD
RS
17 s
core
Kennedy Am J Psych 2011
Data presentedJan 2014 BBRF
Webinar(very optimistic)
New science underway
![Page 11: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/11.jpg)
In Parallel: BROADEN Multi-center RCT SCC DBS for TRD 2008-2014, published 2017
Open, Active allBlinded, active/sham
Part 2: Long Term Follow-up, 2y active DBSPart 1: Randomized blinded 6m; open 6m
15 Centers: 200 planned/90 implanted/4 NR expl<6mStudy halted 2014; data on half of intended sampleAge ≈ 50 (47/90 female)MDD (5 episodes lifetime)Current episode duration ≈ 9-11 yearsPast treatments: 20 lifetime; 8 adequate Txprevious ECT=80%, hosp=80%
11% 24% 40% 51% 47% 10% 20% 30% 27%12% 22% 28% 30%
data on 76 Patients remaining in LTF
Holtzheimer et al. Lancet Psychiatry Oct 2017 Sponsor: St Jude Medical
%Responders %Responders
Progressive change over timeContact changes ∝ improvement
No DTI to verify detailsrole of psychotherapy after 1 year?
Study end: min 2y; range 2-6y, battery Q2yAt study end: Explant or RC offered
Brio #44; Explant #37; Deaths #4; other #5
![Page 12: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/12.jpg)
Science 2013 News Focus
Other Centers, Other Targets, Other LogicOpen label ≠ RCT
Science Focus News 2013
SJMBROADEN
n=90HALTED
MDTRECLAIM
n=30HALTED
+26 AmsterdamRCT 1y D/C
Mt Sinai n=3Shanghai
Germany n=7+16 failed 3m RCTREPEATING NOW
All Targets: ≈ 320 total pts implantedSCC: 162 pts (+ >50 unpublished)VC/VS: 71 pts MFB: > 33 pts (ongoing RCT)
What are we missing?
![Page 13: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/13.jpg)
Binary Public Response to ’Failed’ RCTsimpact on patients and scientists
July 2015
First Question: Is it worth pursuing?
![Page 14: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/14.jpg)
0
25
50
75
100
0 2 4 6 8
Perc
ent o
f Gro
up
Year
Response and Remission Rates
Resp…Remi…
0
5
10
15
20
25
30
0 2 4 6 8
HDRS
-17
IDI
Year
HDRS scores over 8 years
A Crowell
Holtzheimer et al Arch Gen Psych 2012
Blinded Discontinuation Naturalistic discontinuation (battery failure)
Emory Strategy: Follow the Data sustainability; discontinuation, relapse/recapture
N=28 EMORY
P Riva-Posse
A Crowell et al. Am J Psychiatry online Oct 4
![Page 15: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/15.jpg)
How to Reconcile?focus on responder / non-responder differences
1. WHO: patient selection, TRD subtyping.
2. WHERE: target selection, precision targeting
3. WHAT: Readouts of recovery, timecourse
4. HOW: parameter adjustments what/when to maintainClosed loop, on-demand, set-and-forget or fine-tune
Needed at level of individual patientsStart where you can test a null hypothesis
![Page 16: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/16.jpg)
Who: TRD Subtyping TRD patients are NOT homogeneous
Hi negative moodpsychomotor slow
Low positive moodlow motivation
↑SCC25; ↓PM/MCC
↓ventral striatum
medial forebrain bundle
naive past pastAll drug ECT
MCC
SCC FC
Brain biomarker of eligibility? regional abnormalities differ by
Type/number Past Tx failures
FC, FA, PET
SCC FC fMRIWM FA dMRI
CBF PETYIA: K Choi
![Page 17: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/17.jpg)
Considerfull network
not justthe target
diffusion MRI
C25oF11hth
F10
oF11
dACC
hth
vstC25
mF9
Responders Non-Responders
First Clue:Local and
Remote CBFPET changes
2005
Hamani JNS 2008 Riva Posse and Choi Biol Psych 2015
RNR
Location of ActiveContactR vs NR
Toronto Atlanta
NR
R
Where: Are we in the right place?surgical targeting, contact selection, connections
![Page 18: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/18.jpg)
Characterize Common Response ‘Circuit’ necessary and sufficient network not a single region
Voltage Field ModelVolume of Tissue Activated
TAM as seed for DTIUsing specific DBS lead,
WM tracts/locationIndiv stim parameters
Probablistic Tractography
Th
CCnAc
NR to Rw/ contact
changen=5
6 moRespN=6
mF10 ACCmFvst
ThaThvst
Modeled Voxels common to all 6m R same map in all 2y R
impact missingmF and thalamus
Riva Posse and Choi et al Biol Psych 2015
K Choi P Riva-PosseC McIntyre (Case)
Butson & McIntyre Brain Stim 2008
Forcepsminor
Uncinate f.
Cingulum
SCC25
PutativeTemplateFor targeting
![Page 19: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/19.jpg)
Test Benefit of Multipath Targeting Method‘Connectomics’ surgery as concept
Model of Planned VTAStim at predefined location
Forcepsminor
uncinatef.
Cingulum
SCC25
Target Blue-Print
Riva-Posse and Choi et al Molecular Psychiatry 2017
response trajectory
BL 0 1 2 3 4 5 6 9 12
d-DTI in single Ss
8/11R (73%)1. Awake testing in OR2. Chronic DBS at DTI target
w/o contact change over 6m3. single current increase
![Page 20: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/20.jpg)
preop
4w offIntra-optesting
2w on
Atlanta DBS Cohorts 1, 2, 3
time weeks
25
20
15
10
5
HD
RS
17
Forcepsminor
uncinatef.
Cingulum
SCC25
cc
mid-SCC
genu ac
Cohort 1 Cohort 2 Cohort 3
Anatomical target; derived DTI template prospective testing DTI Template Real time DTI
41%73%83%
Arch Gen Psych 2012 Molecular Psych 2017 UH3 in progress 2018
Further Impact of Target Optimizationdiscovery that recovery is not linear
Rate is higher; AND timing is different
Assumption of timeline
![Page 21: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/21.jpg)
Network Reset/Switchacute, rapid
Network Plasticitydelayed, progressive
1 2 3
What ever you just did, I just suddenly shifted …
…I didn’t realize how much work I would need to do myself..
t
E
0
13acute Δ
unstable 2 better well
Chronic ΔProgressive stability
Early reset → remodeling → resilience with time
neg
← M
ood
→ p
os
Sick &Stuck
First DBSChronic DBS
Models to Account for Observed Trajectoryclues to mechanisms; critical for revised study design
Need differential metrics/temporal sampling for different stages?
ketamine
Zarate 2006
![Page 22: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/22.jpg)
Network (Cg25, mF, Ins)Carryover from stim in OR?
PET △ Early PET △ LateNon-network (Lat PF, PCC + pIns)
Change only with active DBS
Cg25
mF10
oF11
MCC
hthbs
Pre 1m 1m 6msham on on
First Toronto Findings: same change pattern
3 and 6 monthsof chronic DBS
Pre 1m 1m 6mSham on on
Need strategy that captures acute changesand progression over time with higher temporal resolution
Evidence of Differential early/late effectsPET CBF changes
Jungho Cha
![Page 23: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/23.jpg)
Why does this matter?(Trial endpoints, treatment adjustments)
Group Mean weekly HDRF ratings
NOW: Use the Same DBS settings for all Phases BUT: Variable response rate in individualsNEED: longitudinal readouts of brain + behavior— relapse vs life stressHYPOTHESIS: different phases show different effects.
individualize to optimize treatment delivery.
preop
4w offcarryover
Intra-optesting
1w onPrimed?
Reset?
2-3m instability
Adjust DBS or Add more therapy?
![Page 24: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/24.jpg)
CBF PET Activa PC+S EGI-hdEEG Pt specificfixed time points ongoing SCC LFP intermittent cortical biophysical Models
self-reported movement autonomic emotion expressionmood, ratings actigraphy, GPS SCR, Heart Rate video, face mm, voice, words
Tracking Chronology of Stimulation EffectsGen-2 devices: causal models, candidate control signals
![Page 25: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/25.jpg)
1
Revisit first exposure to DBS in the ORMonitor patient’s worst symptoms
pain
buriedquicksand
vortex
unrelenting
paralyzed
sticky
voiddisconnected
dead
gnawing
what, when, where change happens?Don’t want to miss potential reset.
![Page 26: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/26.jpg)
I feel more optimistic
I feel more connected
I could wash my hair
I could walk my dog
can imagine seeing friends
DTI, randomized stim130Hz 90us 6mA9 patient: R/L leads8 contacts, 108 trials
I feel lighter
I feel less heavy
I can breathe
the tension is gone
the pain is gone
Characterizing the ‘Depression Switch’pt self report: first evidence of target engagement?
Type 1interoceptive change
Type 2exteroceptive change
30/72 active; 4/36 sham; 17L, 3R 9/72 active (all L); 0 sham
L
mF10
ACC
PCC
mF10/11
L RL
Choi and Riva Posse et alJAMA Neurol Sept 26, 2015
K Choi P Riva-Posse
Forcepsminor
Uncinate f.
Cingulum
SCC25
Type 1: Cingulum Bundle Type 2: CB + Forceps Minor
![Page 27: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/27.jpg)
Cortical Readout of Optimal TargetConfirmation in lab prior to starting DBS
2Hz ERP ON Target based on DTI
Grand Average; n=4, 15 sessions
25ms40ms 70ms
100ms
A Waters
Single SubjectON vs OFF Target
Anatomical specificity
Next step: OR verification
“ON TARGET”CONTACT
ADJACENTCONTACT
Patient A Patient B
“ON TARGET”CONTACT
ADJACENTCONTACT
Waters et al. Human Br Map 2018
![Page 28: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/28.jpg)
L1 2 3 4 5 6 7 8R
baseline
L R B L R B L R B
post CS post OnT
contact sweep repeat On Target
60s avg
1s seg
PSD
Mo SE-Sendi
Elastic-Net Regularization (ENR)
PSD Feature Generator
0 10 20 30
0
0.05
0.1
0.15
0.2
0.25
0.3
Norm
aliz
ed P
SD(z
)
Baseline
Post-OnT
0 10 20 30
0
0.05
0.1
0.15
0.2
0.25
0.3
Norm
aliz
ed P
SD(z
)
Baseline
Post-OnT
↓B
↑A
↓B
Tension is gone, I can move
↑A
-0.15 -0.1 -0.05 0
Normalized Beta Left Change
-0.7
-0.6
-0.5
-0.4
-0.3
Nor
mal
ized
HS
Cha
nge
r=0.6522, corrected p=0.0412, n=9
-2 -1 0 1 2 3 4-1 0 1 2 3 4
Weeks post-opEvidence of carryover. 1-2 weeks
HD
RS
Scor
e
Beta changeTracks with
Carryover effects(I am doing more)
LFP Readout of Depression Switchrepeated bilat DBS at target in OR
![Page 29: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/29.jpg)
A Veerakumar
SCC recording montage
ON OFF
PC+S Chronic recording
Slope and Depression Severity regardless of TIme
1/f slope 𝓧𝓧
14s epochs for analysis
Depressed Responders RemittersMADRS>20 M 11-20 MADRS<11
Med
ian 𝓧𝓧
chan
ge fr
om B
-Pha
se
Left Electrode Right Electrode
Depressed Responders RemittersMADRS>20 M 11-20 MADRS<11
Single Subject Weekly Slope vs Dep Score
B0 B1 C1 C2 C3 C4 C5 C6 E
2.0
1.9
1.8
1.7
1.6
1.5
1.4
1.3
4035302520151050
MAD
RS
right
𝓧𝓧
𝓧𝓧steep - well
𝓧𝓧 flat - sickswitch
DBS off, weekly lab assessment
B Voytek
SCC Weekly Readout first step towards closed loop DBS delivery
Veerakumar et al J Neurophysiology 2019
Slope changed 1 week before Clinical RelapsePutative predictive signal to trigger adjustment
![Page 30: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/30.jpg)
less for tractographyone slide not tworetrospective and prospective
Readouts without Brain or Self-Report? quantify what seems obvious
DBS 35 preop 1 year of DBS
They look different*They move moreThey do thingsThey feel better
Distinguish stalled response, impending relapse, transient life stressor
⊘
⊘
**
*
Rating Scales intuitively less reliable with time
![Page 31: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/31.jpg)
Sahar Harati, Andrea Crowell, HM, Shamim Nemati; IEEE Engineering in Medicine and Biology Society 2016
Tuning DBS based on Facial ExpressionDistinguish depressed vs stressed vs well
video interview n=9; weekly x 6m2min clip, spontaneous speech
S Harati A Crowell
D=T
I=T
D=T
Use toguide doseadjustments
3 Videos @ patientPsychiatrist selected
SickWell
Rough
What does the face say?
![Page 32: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/32.jpg)
Hypothesis:SCC LFP might better track such behavioral
readouts than severity scores
Temp Diff Learning AlgorithmSwitching GLM/Value Iteration
Face-Voice 8-11 wks Predicts 6m OutcomeMin added value of Rating Scale
Behavioral Tracking6 mo Outcome Predictions
Voice print
Faceprint
self-reportRatingscales
S Harati A Crowell
Can you predict when a Patient will Recover?What biomarker best tracks response?
![Page 33: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/33.jpg)
HiImaging
Biophysicalmodels
K Choi A Waters D Obatusin
Voice print
Activity
movement contour mapsdynamics
SteerableNetwork Control
FacialExpression
brainreadouts
Q-Lab at C-ACTquantitative biometrics
S Scherrer D putrino M PhillipsRehab Designer
![Page 34: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/34.jpg)
Toth and…Denison Oxford 2019
Worrell IEEE J Transl Eng Health Med. 2018
MSSM ExperimentsWinter 2019 start
T DenisonOxford
OpenMind
D ObatusinMt Sinai
F Afzal F JamshedOxford
Face + Voice +Time + ratingsInputs to model
B Kopell. M Figee. S Oneill J Gowatsky L Pagan
![Page 35: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/35.jpg)
Neurotechnology and TreatmentEvolution not Revolution
2005-2014
2015
2017
2019
![Page 36: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/36.jpg)
Broken Reset Remodel Rehab/Retrain Relearn Plasticity
• WANT: meaningful symptom relief, sustained, durable (relapse prevention)
• NEED: Rehabilitation strategies that maximize recovery (resilience)
• LEARN: distress ≠ depressed. (Define readouts that can tell the difference)
Recovery takes more than a Stimulatornecessary but not sufficient
Bottom Line: How would you live your lifeif relapse was the exception and not the rule?
![Page 37: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/37.jpg)
What do Patients Think?they get it, but it takes time
I have a lot of learning to do.I sometimes feel quite lost.But it is nothing like before.
I'm just trying to figure out who I am and where I'm headed.I'm somewhat unhappy,
and I'm definitely overwhelmed,but I'm not sick.
Emory #17 (3/10/12)
![Page 38: DBS for Treatment-Resistant Depression: a (5 year ... · Case Western ENTICe Comp Sci Oxford BMI ML/AI Computer Sci Cog NS Biostatistics. modeling modeling Engineering. Clinical](https://reader035.vdocument.in/reader035/viewer/2022081406/5f176079a7149448fd1204bf/html5/thumbnails/38.jpg)
Nash Family Center for Advanced Circuit Therapeuticshttps://icahn.mssm.edu/research/advanced-circuit-therapeutics
For More Information